You are on page 1of 8

Calculation of Thermophysical Properties

of Carbon and Low Alloyed Steels

for Modeling of Solidification Processes

Algorithms and a computer software package for calculating thermophysical material properties
of carbon and low-alloyed steels, associated with the simulation of solidification processes, have
been developed. The earlier studies on kinetic phase transformation modeling are applied and
are the base of the present work. The calculation algorithms are based on thermodynamic theory
connected to thermodynamic assessment data, as well as on regression formulas of experimental
data, and they take into account the temperature, the cooling rate, and the steel composition.
The calculation algorithms and some results of calculations are presented in this article.

I. INTRODUCTION phase. Moreover, these data are usually expressed as a

function of temperature and composition only, i.e., the
THE modeling of solidification systems is a problem cooling rate is not taken into account. So, it is seldom
of a great mathematical and industrial significance. In
possible to find all the data needed. This is especially
recent years, computer simulation models have been de-
the case for carbon and low-alloyed steel grades, be-
veloped to simulate heat transfer in solidification pro-
cause in these steels, even small variations in the com-
cesses. Numerical methods, mostly finite difference or
position may have a significant effect on the
finite element methods, are effectively used. However,
thermophysical material properties. A model that could
to obtain reliable results, accurate data on the thermo-
simulate the thermophysical material properties of steels,
physical material properties are also needed. Typical
taking into account the cooling and the composition,
data needed are the density, the thermal conductivity,
seems not to be available.
and the specific heat. Other important data are the phase
Therefore, algorithms and a computer software pack-
transformation temperatures and the corresponding latent
age for calculating thermophysical material properties of
heats, and also the way in which the latent heats are
steels, taking into account the temperature, the cooling
released during the phase transformations. If so-called
rate, and the composition, have been developed. The
enthalpy formulation is used, the enthalpy values can di-
earlier studies on kinetic phase transformation modeling
rectly be used if they are known. The enthalpy then in-
are applied, and they are the base of the present
cludes all the other data except the thermal conductivity
work. fLz,3] The calculation algorithms are based on
and the density.
thermodynamic theory connected to thermodynamic
The material data are not only functions of the tem-
assessment data and on regression formulas of experi-
perature and the composition but also of the cooling rate.
mental data. The package is now valid for carbon and
This is because the kinetics of phase transformations
low-alloyed steel grades and is under progress for stain-
depend on the cooling rate and the thermophysical prop-
less and alloyed steels. The calculation algorithms and
erties are related to the phases formed. At high temper-
some obtained results are presented in this article.
atures, a high cooling rate typically lowers the phase
transformation temperatures, especially the solidus tem-
perature, but not drastically, due to the compensating II. THERMAL MODELING
effect of a finer dendritic microstructure. At low tem- OF SOLIDIFICATION
peratures, however, the effect of the cooling rate upon
the phase transformation temperatures is much clearer. A general solidification phase change system without
Thus, for accurate simulation of solidification and cool- convection can be described by the following equation:
ing processes, one should be able to know the material
data as a function of temperature, composition, and pc-- =- k + Q [1]
cooling rate. ot Ox
Normally, the material data are taken from the liter-
ature. It is seldom possible, however, to find all the data Here, p is the density, c is the specific heat, k is the
needed. Although material data have been measured for thermal conductivity, and Q is a term describing the rate
a great number of steel grades, most of these data are of energy released by the phase transformations. For so-
valid for the low-temperature region only, and there are lidification phase change, it can be defined as
only few data for higher temperatures up to the liquid
Q = oL o(f~) [2]
JYRKI MIETrlNEN, Research Engineer, and SEPPO Ot
LOUHENK1LPI, Senior Teaching Assistant, are with the Laboratory
of Metallurgy, Helsinki, University of Technology, 02150 Espoo, Here, L is the latent heat of the solidification, and f~ =
Finland. f~(T) is called the solidified fraction in the mushy zone.
Manuscript submitted February 16, 1994. The later term describes the way in which the latent heat


is released during solidification. This way depends are made in one volume element placed on the side of
strongly on the chemical composition of the material to a dendrite arm (Figure 1), assuming a complete mixing
be cast. The solid-state phase transformations can be in the liquid phase. As a result, one obtains the phase
treated in a similar way. Therefore, the material data fractions f*(~b = L, 8, 31) and the phase compositions
needed are the density, the specific heat, the thermal X~ (i = 1 . . . . . n; ~b - L, 3, 3,) as a function of tem-
conductivity, the latent heats, and the phase fractions perature within the range 1000 ~ to 1600 ~ (here, n
during the phase transformations. is the number of components in the alloy). These results,
Equation [ 1] also can be expressed in another form by of course, depend on the cooling rate of the process and
using so-called enthalpy formulation. The enthalpy, H, the chemical analysis of the alloy. Note that the cooling
is defined as the sum of the sensible (fCp. dT) and the rate, given as input data, does not need to be constant.
latent heats. Equation [1] now becomes Normally, the cooling rate varies at different distances
from the casting surfaces and also in time. This can now
p--=-- k [31 be taken into account. Note also that as the model deals
Ot Ox with nonequilibrium solidification, X~ will vary in ferrite
and austenite. Consequently, inside these phases, special
In this case, the necessary material data are the density, concentration profiles will be formed (in liquid, this will
the enthalpy, and the thermal conductivity. The enthalpy not happen due to the complete mixing).
now includes all the other material data, except the den- At lower temperatures (T < 1000 ~ austenite can
sity and the thermal conductivity. Hence, the enthalpy decompose via the following phase transformations: y
values can directly be used if they are known. p s , y --* pc, and y ---->e s + ec, where po~ denotes pro-
In the heat-transfer models, it is usually assumed that eutectoid ferrite, pc denotes proeutectoid cementite, and
the material data are functions of the temperature and e s + ec denotes a mixture of eutectoid ferrite and ce-
the composition only. The cooling rate, however, is also mentite, called pearlite or bainite. In addition, during
an important parameter, because it affects phase trans- rapid cooling, austenite may transform at low tempera-
formations and therefore the thermophysical material tures to carbon-saturated ferrite (martensite) as y ~ m s .
properties. Normally, the cooling rate varies at different In order to simulate these transformations in carbon and
distances from the casting surfaces and also in time. To low-alloyed steels, a semiempirical austenite de-
be precise, this should be taken into account when solv- composition model (ADC) TM was developed. Here, em-
ing the heat-transfer equations. So, for accurate simu- pirical formulas based on continuous cooling
lation of solidification processes, one should be able to
transformation (CCT) experiments were evaluated to
know or model the material data as a function of tem-
predict the temperature ranges of different structures
perature, composition, and cooling rate, and this model (% p s , p c , e s + ec, m s ) as a function of alloy com-
or knowledge should then be coupled in a proper way to
position, cooling rate, and austenite grain size. These
the heat-transfer calculations.
formulas were then coupled to a thermodynamic model,
Equation [1] or [3] is normally solved using fixed
assuming a paraequilibrium condition during the phase
grids. This means that the contraction of the steel is not
transformations. This means that only carbon distributes
calculated and taken into account, i.e., the density
between different phases, whereas the other components
cannot be varied either. In such cases, the density should
be that of the initial liquid and constant. However, tend to inherit the composition of austenite in all phases.
during solidification, the fluid in interdendritic space is As a result of the calculations, one obtains the fractions
free to move, and it more or less compensates the so- of the structures, f ~ (~ = Y, p s , pc, e s + ec, m s ) , and
lidification contractions. To take this feeding into ac- the phase compositions, X~ (i = 1 . . . . . n), as functions
count, the density for the solid phase should be the of temperature within the range T = 25 ~ to 1000 ~
density at the solidus temperature, while above the sol- Note that the cooling rate must be constant in the ADC
idus, the density should vary as a function of model. The reheating or varying cooling rate cannot be
temperature. taken into account. Another problem is the determina-
tion of the original austenite grain size after solidifica-
tion. Accurate data seem not to be available. This item
will be studied more precisely in the future. The IDS
lII. C A L C U L A T I O N O F
Depending on composition, the solidifying steel goes
through some of the following phase transformations:
L --> 8, L --> y, L + 8 --> y, and 8 --> y, where L denotes
liquid, 8 denotes delta ferrite, and y denotes austenite.
In order to simulate these transformations in carbon and
low-alloyed steels, an interdendritic solidification model
(IDS), was developed, ll,21 In this model, the main in-
struments of conventional solute redistribution models
(i.e., material balance equations and Fick's diffusion
laws) are coupled to a thermodynamic solution model,
which relates the compositions of the phase interfaces to Fig, 1 - Longitudinal and transverse cross section of dendrites with
temperature and the phase stabilities. The calculations a volume element on the side of a dendrite arm.


and ADC models are now valid for carbon and low al- In Eq. [11], term G * for liquid, ferrite, and austenite can
loyed steels. be expressed as
When evaluating thermophysical properties for an
alloy, one has to know the properties of the individual
phases and the fractions of these phases. In the case of a : [12]
steels, the phase fractions can be calculated with the IDS
and the ADC models as shown previously. Assuming cb
now that the thermophysical properties are the same for Here, n is the number of components in the alloy, Xi is
all ferritic phases (8, pa, ea, ma) and for both cementite the mole fraction of component i in phase ~b, and/z~ is
phases (pc, ec), the evaluation procedure can be sim- the chemical potential of the component i in that phase.
plified considerably. This implies, however, that the A precise description of the chemical potential depends
total fractions of ferrite and cementite expressed as on the solution model to be used. In this study, the
Wagner-Lupis-Elliott model (WLE) ~4,5"6~ was chosen
f ~ = f a + fpa + f ~ + f,,~ [4] because it can be well applied to the dilute solution
phases of carbon and low-alloyed steels.I2J In addition,
f~ =fPc + f ~ [5] the model is simple and its database can be easily ex-
are known. Here, terms f~, fP'~, f"'~, and fP~ can be cal- tended with special data-conversion formulas, m When
culated directly with the IDS and the ADC models, and applying the WLE model to the phases of iron-base
terms fe,, and f ~ can be approximated from the binary alloys (~b = L, a, y), the chemical potentials of iron and
Fe-C phase diagram as solute elements can be expressed as

(6.69 - pet C)
fea ~ .f~,~+~c [6]
6.69 + RTln X~F~+ GFE~ [13]
pct C
fec .~. .fea+~c [7]
+ (67 '~ - G~) + RT In X~ + Of 6 [14]
where pct C is the carbon content in weight percent and
f~,+e~ is the fraction of pearlite or bainite (or their mix- By means of the terms G~176 - _it-/sERand G~162- G~176
ture) calculated with the ADC model. Therefore, know- i also stands for iron), the Gibbs energy of phase 4, is
ing the fractions and the thermophysical properties of expressed relative to a special reference state (SER), 171
liquid, ferrite, austenite, and cementite, one is able to i.e., the enthalpies of pure elements in their defined ref-
evaluate the properties for the whole steel from room erence phase at 298.15 K and 105 Pa. For iron, the ref-
temperature up to liquid phase. In Sections III-A erence phase is a and for solute elements, it is described
through C, special formulas are given to calculate molar by 0 in Eq. [14]. The third term in Eq. [14], G ~ -
enthalpy, molar specific heat, thermal conductivity, and G ~ , changes the reference state of solute elements in
density for carbon and low-alloyed steels. phase ~b from that of pure elements (Raoultian standard
state) to the state of infinite dilution (Henrian standard
state), in regard to which the effect rising from randomly
A. Calculation of Enthalpy and Enthalpy-Related mixed atoms (term RT In X~) and solutal interactions
Data as Specific and Latent Heats (term G~*) will be expressed. The last term, the partial
According to classical thermodynamics, the molar en- excess Gibbs energy G,-~, is given for iron and the solute
thalpy H and the molar specific heat Cp of a system can elements (i = 2 . . . . . n) as follows: 12,7~
be expressed as n n

H = G - T (;) p
G~ = A" E E RTe.~XiXj + 2B
i=2 j=2

9 RTp'~ + RrpikXiXiXk [15]

Cp = - T \ ~ - 5 ] p [9] i=2 j=2 k=2

n n
~,here G is the molar Gibbs energy of the system. For
heterogeneous phase mixture containing ~r phases, the Gel = E RT~Xj + E RTpJ,X~
nolar Gibbs energy can be given as y=2 y=2
n t/

G=E f4~ [10] +g E [16]

k=l j=2 k>j

vheref ok is the fraction of phase q~k and G ~k is the molar where

3ibbs energy of that phase. For carbon and low-alloyed
teels, Eq. [10] can be written as
A = In 1- X~ + [17]
G = f ' G ~+ f " G '~ +f~'G ~' +f~G r [11] i=2 / i=2


transformations can be calculated in a similar way. The
B = In 1- Xi + X~ specific heat and the latent heats are enthalpy-related

-~ --

X i

X i
)' [ 18]
data, and as mentioned previously, when using the so-
called enthalpy formulation (Eq. [3]), enthalpy values
can be used directly as input data in the heat-transfer
models. In this case, the specific heat and the latent
Here, parameters e~, p~, and p~k describe chemical inter- heats need not be calculated.
actions between different solutes (the phase symbol ~b
has been omitted for the sake of clarity). B. Calculation o f T h e r m a l Conductivity
By ignoring the interactions between the substitutional In carbon and low-alloyed steels, thermal conductivity
solutes in cementite, term G ~ for cementite in Eq. [11] is usually not known for each individual solid phase but
can be approximated as rather for the solid state only. In this case, it is reason-
able to treat the solid phases as a one phase and apply
G~ = -" F:M(G ~ c ~ " ~ "S E R
only two-phase fractions,f ~(liquid fraction) a n d f f (solid
4 M=I,Mr
fraction), in calculations. Knowing that the solid fraction
o ] can be expressed a s f s = 1 - f t , the thermal conductivity
+ 3RT ~ IrMln IFM + ~ I%r [191 of the alloy can be calculated with equation
M = I,M~'2 M=3
k = (1 - f t ) / d + (1 + Amix)ftk/ [21]
Here, Y;a is the site fraction of substitutional element M
occupying sublattice (Fe, M) in cementite (Fe, M)3C. Here,/d and k t are the thermal conductivities in solid and
The term GM~C is the Gibbs energy of pure compound liquid state and A~ix is a parameter describing the effect
M3C, and L~r is a parameter describing the interaction of liquid convection upon the thermal conductivity. If
of Fe and M in sublattice (Fe, M). Note that carbon C the constant Amix is 0, there is no increased heat transfer
designated as 2 occupies another sublattice. By means in the mushy or liquid regions due to convection, i.e.,
0c SER SER 9
of the terms (GM~c -- 3HM -- Hc ), the Gibbs energies
9 the liquid phase is stagnant. This technique, called ef-
of pure compounds M3C are expressed relative to the fective thermal conductivity, is most often used to ac-
reference state SER. count for the convective heat transfer in the liquid during
Values for terms G ~ 1 7 6HSi ER and G ~ - G ~176in solidification, u~ The value of Amix depends on the
Eqs. [13] and [14] have been given by Dinsdale, tTl and mixing intensity, and it can be determined experimen-
values for terms GM~c- 3H sER- HscER and L~r have tally. For a continuous-steel-casting process, for in-
been given by Huang {81 and Qiu. t91 In addition, special stance, a value of Am~x = 4 to 6 may be applied, u~
data-conversion formulas have been derived 121 to evalu- Thermal conductivity has been measured in many
ate values for terms G~6 - G ~ and the interaction pa- carbon and low-alloyed steels at low temperatures, ua-~7j
rameters of Eqs. [ 13] and [ 14]. It should be remembered but at higher temperatures up to the liquidus tempera-
that the published or the evaluated values of all the ture, only a few measurements are available, u4,~6A71
thermodynamic parameters discussed previously are fi- In this work, the thermal conductivity in liquid is as-
sumed to have a constant value of
nally based on experimental measurements such as ac-
tivity measurements, such as activity measurements,
heat-capacity and mixing-enthalpy measurements, and /d = 35
(w) [221
gas equilibrium and solubility data.
When applying Eq. [ 11 ], the ferritic term f " G ~ should
be expressed as f ~ G ~ = fSG~ + fP~C,p~ + f f ~ G e~ + This value, representing the thermal conductivity of pure
f " ~ G "~ because the composition in the ferritic phases is iron at T~, was chosen because of the lack of data for
not the same. The thermodynamic data of ferrite, how- liquid steels. In addition, above the A3 temperature, the
ever, are the same for all the terms G ~, GP~, G ~, and influence of alloying upon the thermal conductivity is
G'% In the case o f cementite, one may apply G ~ = known to be small, f17~
G p~ -- G ~ due to the paraequilibrium condition assumed. In the solid, the thermal conductivity is evaluated by
Finally, it should be remembered that when applying the applying a linear regression analysis to the experimental
equations of this section to calculate the molar enthalpy measurements. ~4"~5"16~ As a result of the analysis, four
or the molar specific heat of the alloy, the mole fraction formulas of the type
X~ in the equations refers to the average composition in
the phase. This can be estimated from the concentration ld = ko + "~a,Ci [23]
profiles obtained with the IDS or the ADC model.
Now, knowing H and Cp by means of Eqs. [8] and
[9], one can then approximate the latent heat of solidi- were obtained corresponding to temperatures 1000 ~
fication (L) from Eq. [20] as 400 ~ 200 ~ and 25 ~ as shown in Table I. In these

L = AHr__,r, - f/ s
Cp. d T [20]
formulas, ai represents the effect of solute i upon/d, and
C, is the composition of solute i. As can be seen, the
correlation coefficients are quite high.
Assuming that kS -= k~ at the liquidus temperature,
where Tz and T~ are the liquidus and the solidus temper- Eq. [22] and the formulas of Table I can be used to inter-
atures of the alloy. The latent heats of solid state phase polate a value for/d at any temperature. At temperatures


Table 1. Thermal Conductivity k ~ ko + Ea,C~ in Low-Alloyed Steels at Different Temperatures*

T (~ ko ac t/si aM, ac, aMo aNi av N r

1000 28.12 -1.60 -0.18 -0.55 . . . . 9 0.93
400 47.70 -7.70 -5.12 -3.37 -3.10 0.59 - 1.68 -6.07 71 0.93
200 57.56 -9.98 -8.85 -6.65 -5.00 -0.48 -3.12 -8.84 73 0.94
25 68.15 -15.9 -12.4 -12.5 -7.87 -1.71 -4.61 -9.99 66 0.95
*The. value of k is given in W/Km and C is given in Weight Percent, N = number of alloys, and r = correlation coefficient.

T < 1000 ~ however, the interpolation should be made In Eq9 [25], the density of liquid, ferrite, austenite,
with the I d / T slope of the previous temperature range and cementite can be calculated from the equations
until proeutectoid ferrite, proeutectoid cementite, or
pearlite begins to form at T = T' (Figure 2). After this, p t = (8319.49 - 0 . 8 3 5 . T )
the interpolation should be made between T --- T' and 9(1 - 0.01 -pct C) + AL [26]
T = 400 ~ by applying the proper/,g values of these
temperatures. p~ = (7875.96 - 0.297 9T - 5.62" 10 -5. T 2)
9(1 - 0 9 pet C) + A s [27]
C. Calculation o f Density p~ = (8099.79 - 0.506. T)
In general, the density of a heterogeneous phase mix- 9(1 - 0.0146- pet C) + A s [28]
ture containing 7r phases can be defined as [~81
pC = 7686 9 - 0.0663.T- 3.12- 10 - 4 - T 2 [29]
p = 1/~f 4,k/ P ~k [24]
A L = - 6 7 . 5 pct Si - 3.9 pct Mn - 8.6 pct Cr
+ 24 pct Mo + 3.3 pct Ni [30]
where f+K is the fraction of phase d~k and p~" is the den-
sity of that phase. In the case of carbon and low-alloyed A s = - 6 3 . 1 pct Si - 6.1 pct Mn - 9.3 pct Cr
steels, Eq. [24] can be expressed as + 2.6 pct Mo - 09 pct Ni [31]
Here, density p is given in k g / m 3, temperature T is
p= 1
/f j +~--~+p +
,= ,j] [251 given in ~ and solute contents are given in weight per-
cent. These sentences, excluding terms A L and A s, are
based on the analysis of Jablonka et al. tZsj for pure iron
where f o and f c are given by Eqs. [4] and [5]. At high and iron-carbon alloys9 The term AL was evaluated from
temperatures (T > 1000 ~ terms f t , f~, and f r , are the density measurements for liquid binary iron
calculated with the IDS model and the o t h e r f '~ terms are alloys, I~9,2~ and the term A s was calculated by applying
set to zero, whereas at low temperatures, terms f / and a linear regression analysis to the density measurements
f~ are set to zero and the other f '~ terms are calculated given for multicomponent steels at the room tempera-
with the ADC model, applying Eqs. [6] and [7]. ture. I~6~ The regression analysis included 30 alloys and
yielded the value 0.987 for the correlation coefficient 9
In the analysis, carbon was not included as a regression
variable, but its effect was taken into account by apply-
ing the following formulation:

pmeasurea= 6 . 6 9 / [ 6 . 6 9 _.C~_dpct
p + pctC]pc

O + Ea,Ci [32]
where the term Y.a,Ci equals the term A s to be solved,
o and p" and pC were calculated from Eqs. [27] and [29]
t- by setting T = 25 ~ and A s = 0. Hence, the A s term

describes the effect of solutes, excluding carbon, upon

i i
the density of ferrite at the room temperature. This term
25 200 400 T' 1000 TL
also is applied to austenite because of the lack of the
Temperature (~ density measurements for that phase.
Fig. 2 - - S c h e m a t i c view on interpolation o f thermal conductivity/(
below the liquidus temperature (T~) by applying known k~ values of IV. EXAMPLES AND DISCUSSION
certain temperatures (black spots). The white spot refers to the for-
mation o f proeutectoid ferrite, proeutectoid cementite, or pearlite at The IDS and A D C models calculate the temperature
temperature T'. ranges of phase transformations, as well as the phase


fractions during the transformations. These models were 1400

tested by comparing the calculated results with numer-

ous experimental values, and a reasonable correlation
was found between the results. ~1,2,3~ The calculation of
thermal conductivity is based on regression formulas of 13J
experimental data and, due to the high values of the cor- ,.,~ 1100
relation coefficients (Table I), the results should be quite
reliable. At high temperatures, however, the data avail- 1000
able are meager and their validity is difficult to estimate.
The calculation of density is based on the results of
the IDS and ADC models and on experimental data. For
9OO \s.~{
pure iron and iron-carbon alloys, the experimental data 800 I t I I I

were taken from the analysis of Jablonka et al. t~81 They 1300 1351] 1400 1450 1501] 1550 1600
calculated the effect of carbon on the density in iron- T (*C)
carbon alloys as a function of temperature and found a
good agreement between the calculated and measured re-
suits. In order to take other solute elements into account,
601] pa ,
regression formulas were evaluated. Enough data to
make an acceptable regression were found, however,
only at 25 "C and in the liquid range. Thus, one equation a~500
was evaluated for the liquid phase and one for 25 ~
The latter also is used over the range between 25 ~ and -r. 400
the solidus temperature. The error is not necessarily sig-
nificant, but one aim for the future work is to find more 300
data for the density. On the other hand, when using fixed
grids in the heat-transfer calculations, the contraction of 200 I I .... t J
the steel is not calculated and taken into account and so 500 600 700 800 900 I000
the density cannot be varied either (Section II).
The calculation of enthalpy and enthalpy-related data T (*C)
as specific and latent heats is based on the results of the Fig. 3 - - E n t h a l p y of pure iron and steel, 1 pct Mn, 0.02 pct P,
IDS and ADC models, thermodynamic theory, and 0.3 pct Si containing 9.1 pct C (steel 1) or 0.6 pet C (steel 2) and
thermodynamic assessment data. The thermodynamic cooled at a rate of T = 1 ~ Symbols are L = liquid, 6 =
delta ferrite, 7 = austenite, S = solidus temperature, p a = proeutec-
data bank of the models contains data for the following toid ferrite, P = pearlite, and P' = end of pearlite reaction_
elements: Fe, A1, B, C, Ca, Cr, Cu, Mn, Mo, N, Nb,
Ni, O, P, S, Si, Ti, and V. Because of the numerous
thermodynamic assessments available today, especially
for iron-base alloys, the thermodynamic calculations can In the case of the 0.6 pct C (steel 2), the solidification
from this point of view be considered quite reliable. An is totally austenitic. At lower temperatures, austenite de-
accurate comparison between calculated and measured composes in both steels to proeutectoid ferrite ( p a l and,
results has been made for pure iron and the Fe-C system later, also to pearlite (P).
only. In both cases, an excellent agreement comparable Figure 3 shows that in steel 1, not only the formation
to those in References 21 and 22 was obtained. For al- of delta ferrite but also the formation of austenite (3,),
loyed steel, only a few experimental data are available, proeutectoid ferrite ( p a l , and pearlite (P) accelerates the
and we are now searching for more data to do accurate heat release. In steel 2, the heat release during the pearl-
comparisons. ite reaction is quite effective due to the quite narrow
Figures 3 through 6 show calculated results for pure temperature range T e to T P" of the austenite de-
iron and for two steel grades, which differ only in carbon composition. Note that in the case of steel 1, this range
content. The effect of the composition on the results is much wider, from T p" to T e'.
clearly can be seen. For instance, in pure iron, the phase Figure 4 shows that proeutectoid ferrite ( p a l increases
transformations take place at constant temperatures, the specific heat of pure iron. At the Curie point of pure
whereas in steels within a certain temperature range. The iron (770 ~ the specific heat increases sharply and
phase transformations taking place are as follows: then decreases smoothly as the temperature decreases.
In steel 1, specific heat increases due to the formation
for pure iron, L ---> 8 ---> 3/--> p a
of proeutetoid ferrite and pearlite. The increase is
for the 0.1 pct Csteel, L ---> ~, L + ~ ----> % ~ ~ 3,,
weaker than in pure iron because of the gradually in-
31 --> pa, 3/----> P (ea + ec) creasing ferrite fraction during the y --> p a trans-
for the 0.6 pct Csteel, L ~ % 3' --> pa, formation. In the case of steel 2, the strong increase of
3, ~ P (ea + ec) the specific heat due to the pearlite formation can be
In the case of iron and O. 1 pct C steel (steel 1), the explained by the narrow temperature range of austenite
first phase to be formed is delta ferrite (~). Some degrees decomposition.
above the solidus (S), steel goes through the peritectic The cooling rate is also an important parameter. While
transformation (L + t$ ---> 3,), which then, below the sol- affecting the kinetics of phase transformations, it also
idus, changes to the solid-state transformation (~ ---> 3,). affects the thermophysical properties of the alloy. As an


0.9 9 -'---7 7500


v 0.8 7mo
"" 7200
o. _e.
0 0.7
y:~5 s

0.6 : i + + i
69OO t t 9 I I
1301) 1351) 1400 1450 IS(X) 1550 1600 1300 1~ 1400 1450 1500 1550 1600
t (*C) T (~
i.1 7800

.~ p. r
v-'- 0.9 E 7700

o. Z-
0 0.7 76oo

0.5 : ~ ~ - 7500 + - I I I
5OO 60O 7OO 800 900 !000 500 600 700 800 900 I000
T (~ T (*(2:)

Fig. 4 - - S p e c i l i c heat of pure iron and steel, 1 pct Mn, 0.02 pet P, Fig. 6 - - D e n s i t y of pure iron and steel, 1 pet Mn, 0.02 pct P,
0.3 pct Si containing 9-1 pet C (steel 1) or 0.6 pct C (steel 2) and 0.3 pct Si containing 0..1 pct C (steel 1) or 0.6 pct C (steel 2)
cooled at a rate of T = 1 ~ Symbols are L = liquid, (5 = and cooled at a rate o f t = 1 ~ Symbols are L = liquid, 6 =
delta ferrite, y = austenite, S = solidus temperature, pc~ = pro- delta ferrite, y = austenite, S = solidus temperature, pot = proeutec-
eutectoid ferrite, P ~ pearlite, and P ' = end of pearlite reaction. told ferrite, P = pearlite, and P' = end of pearlite reaction.


30 0.6

2S P
0.5 ~ ~ i i I
500 550 600 650 700 7.50 800
0 200 400 600 800 ! 000 ! 200 1400 1600
T (~
T (*C)
Fig. 7 - - S p e c i f i c heat of steel, 0.2 pet C., I pet Mn, 0.02 pc.t P,
Fig. 5 - - T h e r m a l conductivity of pure iron and steel 1 pet Mn, 0.3 pct Si cooled at a rates of (a) T = 10 ~ (b) T =
0.02 pct P, 0.3 pct Si containing 0.1 pct C (steel I) and 0.6 pct C 0.1 ~ and (c) T = i0 ~ at T > 1000 ~ and T = 0.1 ~ at
(steel 2) and cooled at a rate of T = 1 ~ Symbols are: pot = T < 1000 ~ Symbols are pot = proeutectoid fertile, P = pearlite,
proeutectoid ferrite and P = pearlite. and P' = end of pearlite reaction,

example, Figure 7 shows the effect of cooling rate upon also that a high cooling rate during solidification refines
the specific heat at lower temperatures. As can be seen, the grain size. This also affects the kinetics of the phase
a high cooling rate delays the formation of new phases transformations and the material properties (case c in
~cases a and b). The cooling rate also affects the frac- Figure 7).
:ions of phases to be formed during phase transforma- Figure 8 shows the calculated latent heat values of so-
;ions and, also in this way, the material properties. Note lidification for the iron-carbon system and for a steel


lation algorithms are based on thermodynamic theory
connected to thermodynamic assessment data, as well as
on regression formulas of experimental data. The pack-
0) 250 ----- - Pet~oeffe ": . . . . . . . . . . . .
age is now valid for carbon and low-alloyed steel grades
and is under progress for stainless and alloyed steels.
..a 245 Some comparisons between the calculated results and
measured values were done and a good agreement was
24O found, l~,z,31 but more comparisons are planned, espe-
L'->ferrite,~ Auslenite formafiQn starts cially on enthalpy and on enthalpy-related data, such as
235 ,
specific and latent heats. The development work of the
0 O. 1 0.2 0.3 0.4 0.5 0.6 calculation algorithms will also be continued.
Fig. 8--Calculated latent heat of solidification as a function of ACKNOWLEDGMENT
carbon content for Fe-C system and for steel A: X pct C, 1 pct Mn,
0.02 pct P, and 0.3 pct Si, cooled at a rate of ~" = 1 ~ The authors wish to thank Professor Lauri Holappa
with the Helsinki University of Technology and the
Technology Development Center for financial support.
grade A, 1 pct Mn, 0.02 pct P, 0.3 pct Si, as a function
of carbon content. Up to about 0.1 pet C, the liquid so-
lidifies fully to ~-ferrite and the latent heat decreases REFERENCES
with increasing carbon content (or with decreasing sol-
1. J. Miettinen: Metall. Trans. A, 1992, vol. 23A, pp. 1155-70.
idus temperature). Above about 0.1 pct C, the steels so- 2. J. Miettinen: Ph.D. Thesis, Helsinki University of Technology,
lidify partly to austenite through the peritectic reaction Espoo, Finland, 1992.
and, with increasing austenite fraction, the latent heat 3. J. Miettinen: 1-lelsinki University of Technology, Espoo, Finland,
increases. At the peritectic composition (for the Fe-C- unpublished research, 1994.
4. C. Wagner: Thermodynamics of Alloys, Addison-Wesley,
system 0.17 pet C), the steel already solidifies fully to Reading, MA, 1952.
austenite (through the peritectic reaction). The latent 5. C.H.P. Lupis and J.F. EllioU: Acta Metall., 1966, vol. 14,
heat is highest at the peritectic composition. Above the pp. 529-38.
peritectic composition, the latent heat again decreases 6. S. Srikanth and K.T. Jacob: Metall. Trans. B, 1988, vol. 19B,
with increasing carbon content (or with decreasing sol- pp. 269-75.
7. A.T. Dinsdale: CALPHAD, 1991, vol. 15, pp. 317-425.
idus temperature). 8. W. Huang: Metall. Trans. A, 1991, vol. 22A, pp. 1911-20.
The calculation package is now valid for carbon and 9. C. Qiu: Ph.D. Thesis, Royal Institute of Technology, Stockholm,
low-alloyed steel grades and is under progress for stain- 1993.
less and alloyed steels. The development work of the 10. J.E. Lait, J.K. Brimacombe, and F. Weinberg: lronmaking and
calculation algorithms will also be continued. One ques- Steelmaking, 1974, vol. 1, pp. 90-97.
11. E.A. Mizikar: Trans. TMS-AIME, 1967, vol. 239, pp. 1747-53.
tion, for instance, is what the original austenite grain 12. I. Saucedo, J. Beech, and G.J. Davies: Met. Technol., 1982,
size is after solidification. Accurate data seem not to be vol. 9, pp. 282-91.
available. Moreover, additional experimental CCT data 13. S. Louhenkilpi, E. Laitinen, and R. Nieminen: Metall. Trans. B,
will be used for the determination of the regression for- 1993, vol. 24B, pp. 685-93.
14. F. Richter: Die Wichtigsten Physikalischen Eigenschaften yon 52
mulas in the ADC model, and more thermodynamic Eisenwerkstoffen, Verlag Stahleisen M.B.H., Dusseldorf,
assessment data will be added to the database of the Germany, 1973.
model package. More data on thermophysical properties 15. F. Richter: Physikalischen Eigenschaften yon Stiihlen un ihre
are searched and comparisons between calculated values Temperaturabhiingigkeit, Verlag Stahleisen M.B.H., Dusseldorf,
Germany, 1983.
and experimental results will be continued. One future 16. C.D. Smithells: Metals Reference Book, 5th ed., Butterworth and
aim also involves connecting the calculation package it- Co. Ltd., London, 1976.
eratively to a heat-transfer model of continuous steel 17. P.J. Wray: Proc. Modeling of Casting and Welding Processes,
casting. Rindge, NH, Aug. 3-8, 1980, H.D. Brody and D. Apelian, eds.,
TMS-AIME, Warrendale, PA, 1981, pp. 245-57.
18. A. Jablonka, K. Harste, and K. Schwerdtfeger: Steel Res., 1991,
V. CONCLUSIONS vol. 62, pp. 24-33.
19. C. Benedicks, N. Ericsson, and G. Ericsson: Arch.
Algorithms and a computer software package for cal- Eisenhiittenwes., 1930, vol. 3, p. 473.
20. A. O|sson: Scand. J. Metall., 1981, vol. 10, pp. 263-71.
culating thermophysical material properties of steels, 21. A. Fernandez Guillermet and P. Gustafson: High Temp.-High
taking into account the temperature, the cooling rate, Pressures, 1985, vol. 16, pp. 591-610.
and the composition, have been developed. The calcu- 22. P. Gustafson: Scand. J. Metall., 1985, vol. 14, pp. 259-67.